An algorithm for mean pyramids vector quantization using Hadamard transform and tree structure

Compressing an image is a significant technique for the sake of saving a little more bandwidth or storage space. However, one of the main problems in vector quantization (VQ) is the high computational complexity of searching for the appropriate codeword in its codebook. To overcome this problem a fast search algorithm based on Tree-Hadamard Transform of the mean pyramid levels of the codewords is proposed. The algorithm uses tree structure of the highest level for all codewords to speed up the searching process. The proposed algorithm is more efficient than using only Hadamard transformed mean pyramids especially in high dimension cases.